Gas hydrate is commonly found concentrated in sand- and silt-rich reservoirs and these are thought to have great potential as an energy resource. To evaluate this potential, it is critical to understand their in-situ stress state and their behavior during production. We explore the geomechanical behavior of sandy silt sediments from a hydrate reservoir at Green Canyon GC 955, deep-water Gulf of Mexico without hydrate and compare these results to previous work on the same sandy silty sediments with hydrate present. Previous results from hydrate-bearing sediments at GC 955 using intact specimens show that the ratio of lateral to axial effective stress under uniaxial strain (K0) is strongly time-dependent and over long timescales approaches 1 (isostatic stress state). New results from sediments without hydrate using reconstituted specimens reveal that K0 is 0.51 and independent of time and stress. The reservoir without hydrate is approximately three times more compressible than with hydrate. In addition, we find that hydrate-free sediments have a normalized creep rate of Cα/Cc of about 0.028 at all stress levels, whereas hydrate bearing sediments exhibit a decrease in Cα/Cc with stress. We use these observations to present a new geomechanical model that describes how stress and porosity evolve during production of the hydrate reservoir. The key insight in our proposed model is that the sediment skeleton and gas hydrate phases share any applied external load in proportion to their relative stiffnesses. The model results predict that hydrate-bearing sediments are stiffer than gas hydrate-free sediments and that K0 reduces dramatically after gas hydrate dissociation, as is observed. At GC 955, a 5 MPa depressurization is needed to cause gas hydrate dissociation. This change in pore pressure increases the vertical and horizontal effective stresses from 3.8 to 8.8 MPa and 3.8–4.4 MPa, respectively. The dramatic vertical effective stress increase will cause significant compaction (7%) and perturb stresses near the wellbore. These experimental observations and the load-sharing model presented have the potential to underpin a new generation of hydrate reservoir simulation models, which may drive more effective production approaches.